Alexander Haw¶

This is a system of devices and an accompanying protocol for transferring files up to a kilometer away. My teammates and I developed this novel implementation of a common idea using several popular technologies including: Python, Raspberry Pi, LoRa, CAD, and FDM.

Video Presentation

Before my time at ABB Napier, we had aquired several thousand large elecrolytic capcitors with a combined value of about $500,000. These were being consumed at a rate much lower than expected when they were purchased. This resulted in the capacitors staying in storage for upwards of two years - which was the manufacturer’s stated shelf life. They consequently manifested unacceptable leakage current once installed, leading to complications at the end-of-line test.

To remedy this and recover the value of the capacitors, I designed and (with the assistance of our test technicians) assembled a reforming fixture. This was ultimetely nothing but an apparatus that slowly charged and capacitors to their rated voltagem, then subsequently discharged them to allow safe storage.

Other tasks involved in this project included: coordinating with fabrication services to create custom parts, coordinating material movement and storage requirements with the warehouse staff, reviewing the safety features of the fixtures with R&D engineers (earth bonding for metal parts, isolation of energized components, PPE and measurement tool requirements, Job Safety Analysis form for first test run), documenting the design process, and composing an operating procedure.

Ultimately, this project was able to recoup almost all potential lost value; materials, labor, and transportation amounting to about 5% of the combined parts’ value.

At Zero Motorcycles, our vehicles utilized an ARM-based microcontroller for both the MBB and BMU. However, the microcontroller we used was being discontinued. This presented the following requirement in our manufacturing process: the ability to identify which unprogrammed IC was inside an ECU, then program and configure the ECU accordingly - all without the program user’s guidance. This would allow us to use up our existing stock of ECU’s then seamlessly introduce the new part without any manual intervention.

I developed a Python program for this, capable of identifying the IC via JTAG by reading its ID registers. It then flashed the correct firmware file based on this information, also using the JTAG interface. This poka-yoke was nescessary to eliminate any possibility of a mislabelled ECU recieving the incorrect firmware, leading to unexpected behavior. Once the ECU was programmed however, updates could be applied using the CAN interface.

Considerations were made for ease-of-use in the operator interface, utilizing barcode scanners instead of manual data entry, and traceability included by recording logs for each programming performed.

The Duluth Maker space had recently installed a new CNC router in their wood-shop. In order to operate it however, they needed an electrical schematic for the city to approve. I was referred to this task by my friend Josh from APA. See below for the final product, made in Libre Office. I had a lot of fun working with the DMS staff on this!

In my sophomore year at UMD, I bought a 1977 Chevrolet El Camino as my daily driver. None of the dashboard insturments worked, so I set out to make a replacement. I stripped all of the internals from the cabin, and over a number of weeks created the final product to the left. I designed and fabricated everything you see from scratch.

The EE Lounge is where I spent much of my time between classes, so I made an effort to make it into an inviting place for learning and collaboration. There is a display case next to the entrance, which initially contained a monitor that had long-since stopped working. I took an abandoned Tesla Coil that was once a capstone project and replaced the monitor with it. I then built a Faraday Cage around the coil for safety, then added some buttons so that visitors can energize the coil themselves!

One of the most common grievances from EE undergrads at UMD is the lack of resources available to student for personal projects. This follows from the second most common comment: that there is a lack of hands on teaching in the curriculum. To help with this, I converted one of the UMD student areas to a workspace. This includes a workbench with tools for creating electronics, including a 3D printer.

I created a device at APA to test angle sensors that use the CANopen protocol. These sensors are used in the junction boxes of one of our customers that produce heavy industrial equipment. This is based off the MCP2515 chip, coordinated by an Arduino Uno. This is used by our production team to create a test report to verify that the device is functioning properly upon leaving the plant.

At APA, process instructions are typically written using a word processor. We recieved a large number of schematics for cables from a customer. These have very few components, in contrast to a typical wiring harness. I found that the amount of engineering work required to create process instructions for such a great number of practically identical assemblies was woefully inefficient, where every global revision could take several hours to update. This could be as simple as changing the length of a piece of heat-shrink tube. To make this operation easier, faster, and less susceptible to human error, I created a Visual Basic program that generates HTML documents from data stored in a spreadsheet. These documents are then served to the production staff from a Rasperry Pi running an Apache server. This reduced the workload on the engineering team several fold for this project.

The APA engineering team composes sometimes hundreds of ECOs a day, which are synced to our ERP system following review. These are created by hand, and this led to an endless stream of typos and other various goofs that created an enormous bottleneck in the ECO integration process. I created a program that can by run at the click of a button that checks a long list of common errors in these ECO and alerts the engineer before it becomes a problem!